The reference to any prior art in this specification is not, and should not be taken as, an acknowledgement or any form of suggestion that the prior art forms part of the common general knowledge.
For the purposes of automatic identification, a product item is commonly identified by a 12-digit Universal Product Code (UPC), encoded machine-readably in the form of a printed bar code. The most common UPC numbering system incorporates a 5-digit manufacturer number and a 5-digit item number. Because of its limited precision, a UPC is used to identify a class of product rather than an individual product item. The Uniform Code Council and EAN International define and administer the UPC and related codes as subsets of the 14-digit Global Trade Item Number (GTIN).
Within supply chain management, there is considerable interest in expanding or replacing the UPC scheme to allow individual product items to be uniquely identified and thereby tracked. Individual item tagging can reduce “shrinkage” due to lost, stolen or spoiled goods, improve the efficiency of demand-driven manufacturing and supply, facilitate the profiling of product usage, and improve the customer experience.
There are two main contenders for individual item tagging: visible two-dimensional bar codes, and radio frequency identification (RFID) tags.
There are a significant number of different bar code symbologies, which allow data to be encoded as 2D visible markings, and these include, for example:                “Code 49” described in U.S. Pat. No. 4,794,239;        “Data Matrix” described in U.S. Pat. No. 4,939,354, U.S. Pat. No. 5,053,609 and U.S. Pat. No. 5,124,536;        “Datastrip Code”, as described for example in U.S. Pat. No. 4,692,603, U.S. Pat. No. 4,728,783, U.S. Pat. No. 4,754,127, and U.S. Pat. No. 4,782,221;        “hueCode” described in U.S. Pat. No. 5,369,261 and U.S. Pat. No. 5,118,369;        “Maxicode” described in U.S. Pat. No. 4,874,936, U.S. Pat. No. 4,896,029 and U.S. Pat. No. 4,998,010;        “MiniCode” described in U.S. Pat. No. 5,153,418, U.S. Pat. No. 5,189,292 and U.S. Pat. No. 5,223,701; and,        “PDF 417” described in U.S. Pat. No. 5,243,655.        
Bar codes have the advantage of being inexpensive, but require optical line-of-sight for reading and in some cases appropriate orientation of the bar code relative to the sensor. Additionally they often detract from the appearance of the product label or packaging. Finally, damage to even a relatively minor portion of the bar code can prevent successful detection and interpretation of the bar code.
RFID tags have the advantage of supporting omnidirectional reading, but are comparatively expensive. Additionally, the presence of metal or liquid can seriously interfere with RFID tag performance, undermining the omnidirectional reading advantage. Passive (reader-powered) RFID tags are projected to be priced at 10 cents each in multi-million quantities by the end of 2003, and at 5 cents each soon thereafter, but this still falls short of the sub-one-cent industry target for low-price items such as grocery. The read-only nature of most optical tags has been cited as a disadvantage, since status changes cannot be written to a tag as an item progresses through the supply chain. However, this disadvantage is mitigated by the fact that a read-only tag can refer to information maintained dynamically on a network.
The Massachusetts Institute of Technology (MIT) Auto-ID Center has developed a standard for a 96-bit Electronic Product Code (EPC), coupled with an Internet-based Object Name Service (ONS) and a Product Markup Language (PML). Once an EPC is scanned or otherwise obtained, it is used to look up, possibly via the ONS, matching product information portably encoded in PML. The EPC consists of an 8-bit header, a 28-bit EPC manager, a 24-bit object class, and a 36-bit serial number. For a detailed description of the EPC, refer to Brock, D. L., The Electronic Product Code (EPC), MIT Auto-ID Center (January 2001), the contents of which are herein incorporated by cross-reference. The Auto-ID Center has defined a mapping of the GTIN onto the EPC to demonstrate compatibility between the EPC and current practices Brock, D. L., Integrating the Electronic Product Code (EPC) and the Global Trade Item Number (GTIN), MIT Auto-ID Center (November 2001), the contents of which are herein incorporated by cross-reference. The EPC is administered by EPCglobal, an EAN-UCC joint venture.
EPCs are technology-neutral and can be encoded and carried in many forms. The Auto-ID Center strongly advocates the use of low-cost passive RFID tags to carry EPCs, and has defined a 64-bit version of the EPC to allow the cost of RFID tags to be minimized in the short term. For detailed description of low-cost RFID tag characteristics, refer to Sarma, S., Towards the 5c Tag, MIT Auto-ID Center (November 2001), the contents of which are herein incorporated by cross-reference. For a description of a commercially-available low-cost passive RFID tag, refer to 915 MHz RFID Tag, Alien Technology (2002), the contents of which are herein incorporated by cross-reference. For detailed description of the 64-bit EPC, refer to Brock, D. L., The Compact Electronic Product Code, MIT Auto-ID Center (November 2001), the contents of which are herein incorporated by cross-reference.
EPCs are intended not just for unique item-level tagging and tracking, but also for case-level and pallet-level tagging, and for tagging of other logistic units of shipping and transportation such as containers and trucks. The distributed PML database records dynamic relationships between items and higher-level containers in the packaging, shipping and transportation hierarchy. IBM Business Consulting Services, in conjunction with the Auto-ID Center, has carried out a number of case studies analysing and quantifying the costs and benefits of RFID-carried EPCs in the supply chain. They distinguish the benefits which accrue at different stages in the supply chain (e.g. distribution versus retail), at different levels of tagging (i.e. pallet versus case versus item), in response to different sources of loss (e.g. shrinkage versus unsaleables), and across different product categories (e.g. grocery versus apparel versus consumer electronics). Since the Auto-ID Center exclusively advocates RFID-carried EPCs, the case studies do not clearly distinguish the benefits which accrue from EPCs alone from the benefits which accrue specifically from RFID tags. In addition, the case studies implicitly adopt a very optimistic view of the omni-directional scanning performance of RFID in the presence of radiopaque product, i.e. typically liquid content and metal packaging. More broadly, the case studies do not clearly recognise benefits already beginning to accrue from systemic supply chain changes, such as better utilisation of UPC-based scan data collected at the point-of-sale, increasingly automated reordering and replenishment, and improving levels of communication and data sharing between different participants in the supply chain. In many cases these changes are presented as if predicated on Auto-ID technologies such as RFID-carried EPCs, when in fact they are not. This in turn tends to overstate the benefits of these technologies.
The case studies implicitly assume that tagged units can be accurately scanned in bulk, e.g. when a pallet-load of tagged cases is moved within a distribution center. However, a study by Alien Technology, the first manufacturer of RFID tags conforming to the Auto-ID Center's UHF RFID specifications, shows that cases of radiopaque product (such as soft drinks, shampoo, detergent, and coffee in metal containers) can only be effectively scanned when the case tags are within line-of-sight of tag readers as discussed in Alien Technology, “RFID Supply Chain Applications—Building Test 1”, February 2002. In practice this means that pallets of radiopaque product must be split so that individual cases can be conveyed past tag readers, precluding pallet-level operations including storage and dock-to-dock transfer.
Although not directly explored in the Alien study, the same restrictions apply at the item level. For example, while the case study on obsolescence Alexander, K. et al., Applying Auto-ID to Reduce Losses Associated with Product Obsolescence, MIT Auto-ID Center, November 2002, assumes that shelf scanners in a retail store can perform a complete scan of shelf stock, and the case study on shrinkage Alexander, K. et al., Applying Auto-ID to Reduce Losses Associated with Shrink, MIT Auto-ID Center, November 2002, assumes that exit scanners in a retail store can successfully read items jumbled together in a shopping cart or in grocery bags, in reality the presence of radiopaque product is likely to undermine performance in these situations, thereby compromising some of the claimed benefits of RFID. The Auto-ID Center's own study of supermarket shelf reader design factors concludes that UHF radiopaque product items should have their RFID tags attached to their tops within line-of-sight of shelf readers Cole, P., A Study of Factors Affecting the Design of EPC Antennas & Readers for Supermarket Shelves, MIT Auto-ID Center, 1 Jun. 2002.
As with case-level RFID scanning in the distribution center, item-level RFID scanning in the retail store works best when items are handled individually, such as during stock movement to and from shelves, and during the checkout process, i.e. where each item is allowed to fall within line-of-sight of the reader.
The case studies generally conclude that benefits accrue predominantly from case-level tagging when the case is the primary unit of product movement, which remains true right through the supply chain to the retail store backroom.
Benefits from item-level tagging begin to accrue in the retail store once cases are split and product hits the shelves, and these benefits fall into three main categories: a reduced shrinkage rate; a reduced unsaleable rate; and reduced out-of-stocks (with less safety stock). These benefits are discussed in detail below.
Stage-relevant tagging levels are illustrated in FIG. 100.
The case studies assume seven product categories, summarised in Table 1. For every product category except grocery, the case studies conclude that item-level tagging is cost-effective. Specifically, the case studies do not consider item-level RFID tagging in grocery to be cost-effective because of the high cost of RFID tags relative to the average item price.
Note that if partial and incremental item-level RFID tagging of higher-value grocery items occurs (such as of packets of razor blades Alien Technology, “Alien Announces Major Order for Low-cost RFID Tags”, 6 Jan. 2003, http://www.alientechnology.com/library/pr/alien_gillette.htm, then from the point of view of per-tag cost it becomes more difficult to justify item-level tagging of remaining products, since the average price of untagged items has been reduced. Conversely, it may become easier to justify from the point of view of sunk investment in reader infrastructure.
TABLE 1Product categories and average item pricesproduct categoryaverage item priceGrocery  $1.75Apparel$14consumer electronics$130 health & beauty $9music & video$18Pharmacy$27Toys$18
The case studies therefore make a convincing argument for case-level RFID tagging for all product categories. Additionally item-level RFID tagging may be used for more expensive items.
With item-level tagging, each product item is assigned a unique EPC at time of manufacture. The item's EPC then serves as a key into a distributed PML database which records the characteristics of the item and its evolving history as it proceeds through the supply chain. This includes the item's inclusion in a dynamic hierarchy of packaging, shipping and transportation units, each identified by its own unique EPC. Tracking of higher-level units through the supply chain implicitly support the tracking of lower-level units. For example, once a pallet is loaded and until it is unloaded and split, pallet-level tracking is sufficient to also track its case-level content. Similarly, once a carton is filled and until it is re-opened and split, case-level tracking is sufficient to also track its item-level content. Readers installed in entry and exit portals in factories, warehouses, distribution centers and retail stores can automatically track unit movements and update movement histories. Notwithstanding issues with automatically tracking radiopaque product, RFID readers have benefits for pallet-level and case-level tracking.
At the checkout, the unique EPC of the item prevents it from being recorded as a sale more than once. This allows the checkout to be partially or fully automated. Automatic scanning of a traditional UPC bar code, which only identifies item class, is problematic because multiple scans of the same item are difficult to avoid and impossible to detect from the bar code alone. In an automatic checkout the EPC of an item is typically read many times to ensure that the EPC is read at all, but is only recorded as a sale once. The unique EPC also prevents the checkout operator from multi-scanning a single item to account for a number of similar items, a common time-saving practice which can lead to inventory inaccuracy and thereby undermine automatic reordering and replenishment.
It has been suggested that an RFID-based automatic checkout process can be as simple as wheeling a shopping cart full of RFID-tagged product items through a checkout zone continuously scanned by one or more RFID readers.
In reality, due to issues with radiopaque grocery items, an RFID-based automatic checkout is likely to require each item to pass through the RFID reader's field individually. This may happen when the customer places the item in the cart, i.e. if the cart incorporates a reader, but is more likely to happen at the checkout where the operator or customer either places each item on a conveyor to transport the item through the reader's field, or manually presents each item to the reader's field.
Similarly, whilst the use of item-level RFID tagging arguably makes it possible to construct so-called smart shelves which incorporate RFID readers and continuously monitor RFID-tagged shelf content, practically this is once again subject to performance in the presence of radiopaque product.
The cost of the RFID tag approach is particularly of importance in the grocery sector which is characterised by high-volume sales of low-priced product items, coupled with low net margins. In 2001-2002 the United States grocery sector achieved net profits of 1.36% on net sales of roughly $500 billion.
During the same period the grocery sector experienced a shrinkage rate of 1.42% and an unsaleable rate of 0.95% Lightburn, A., 2002 Unsaleables Benchmark Report, Joint Industry Unsaleables Steering Committee 2002. Net profit and shrinkage were therefore roughly equal at $7 billion each, and unsaleables accounted for an additional $5 billion. Out-of-stocks were further estimated to result in a 3% loss in net sales Grocery Manufacturers of America (GMA), Full-Shelf Satisfaction—Reducing Out-of-Stocks in the Grocery Channel (Executive Summary), 2002, which translates into a $200 million reduction in net profit. The grocery sector is also highly labour-intensive, with labour costs accounting for more than 50% of operating expenses. Profitable operation in the grocery sector therefore relies on maximising efficiency, minimising losses due to shrinkage, minimising losses due to unsaleables, and minimising out-of-stocks while minimising levels of safety stock. Table 2 summarises these sources of loss in the grocery sector.
TABLE 2Sources of loss in the grocery sectorApproximate costsource of losscontribution($millions)Shrinkage1.42%7,000Unsaleables0.95%5,000out-of-stocks0.04%204Total2.41%12,204
The grocery sector is likely to significantly reduce these sources of loss over the coming decade, independently of item-level tagging, by better utilising UPC-based scan data collected at the point-of-sale, by increasing the level of automation of reordering and replenishment, and by improving communication between different participants in the supply chain. Furthermore, the benefits of item-level tagging itself only accrue if such systemic changes actually take place. However, the cost of apply RFID tags to provide item level tagging to further enhance loss reduction is currently cost prohibitive.
As shown in Table 2, the cost of shrinkage, unsaleables and out-of-stocks amounts to about 2.41% of net sales. Assuming an average grocery item price of $1.75, this cost equates to about 4.2 cents. Further assuming universal tagging of grocery items, and ignoring other costs and benefits of item-level tagging, such as the cost of the reader infrastructure and the benefit of an improved consumer experience, 4.2 cents represents an absolute upper limit on the threshold cost of a tag in the grocery sector.
The Auto-ID Center hopes to achieve a 5 cent EPC-compatible passive RFID tag within the next couple of years, and Alien Technology are moving towards that goal with a tag design which they expect to price at 10 cents in multi-billion tag volumes by the end of 2003. Alien Technology, 915 MHz RFID Tag, .Ghassali, M., Unsaleables “The U.S. Experience”, Unilever Bestfoods North America, 27 Mar. 2001. However, the 5 cent tag goal is still highly speculative, and even in multi-billion tag volumes there is currently no projected timeline for achieving an RFID tag price lower than 5 cents. Despite this, the IBM Auto-ID case studies assume a two cent RFID tag within a couple of years in their most optimistic scenarios Alexander, K. et al., Applying Auto-ID to Reduce Losses Associated with Shrink, MIT Auto-ID Center, November 2002.
Since even wildly optimistic projected cost savings only marginally justify the cost of the most optimistically-priced RFID tags, it is unlikely that universal item-level RFID tagging in the grocery sector is justified in the foreseeable future.
In addition to this however, other disadvantages of the RFID tagging scheme, such as the difficulty of scanning in the presence of radiopaque products, and issues surrounding privacy, make the use of RFID tags undesirable in item-level tagging of more expensive products even where the RFID cost becomes negligible.
It is therefore desirable to find an alternative to the use of RFID tags for item level tagging which ensures reliable item identification, which does not suffer from drawbacks such as reduced privacy for the consumer. It is also preferable that the technique provides a lower cost alternative thereby allowing it to be economically used on grocery items.